Identification of Dioxygenases Required for Aspergillus Development

Aspergillus sp. contain ppoA, ppoB, and ppoC genes, which code for fatty acid oxygenases with homology to fungal linoleate 7,8-diol synthases (7,8-LDS) and cyclooxygenases. Our objective was to identify these enzymes, as ppo gene replacements show critical developmental aberrancies in sporulation and pathogenicity in the human pathogen Aspergillus fumigatus and the genetic model Aspergillus nidulans. The PpoAs of A. fumigatus and A. nidulans were identified as (8R)-dioxygenases with hydroperoxide isomerase activity, designated 5,8-LDS. 5,8-LDS transformed 18:2n-6 to (8R)-hydroperoxyoctadecadienoic acid ((8R)-HPODE) and (5S,8R)-dihydroxy-9Z,12Z-octadecadienoic acid ((5S,8R)-DiHODE). We also detected 8,11-LDS in A. fumigatus and (10R)-dioxygenases in both Aspergilli. The diol synthases oxidized [(8R)-2H]18:2n-6 to (8R)-HPODE with retention of the deuterium label, suggesting antarafacial hydrogen abstraction and insertion of molecular oxygen. Experiments with stereospecifically deuterated 18:2n-6 showed that (8R)-HPODE was isomerized by 5,8- and 8,11-LDS to (5S,8R)-DiHODE and to (8R,11S)-dihydroxy-9Z,12Z-octadecadienoic acid, respectively, by suprafacial hydrogen abstraction and oxygen insertion at C-5 and C-11. PpoCs were identified as (10R)-dioxygenases, which catalyzed abstraction of the pro-S hydrogen at C-8 of 18:2n-6, double bond migration, and antafacial insertion of molecular oxygen with formation of (10R)-hydroxy-8E,12Z-hydroperoxyoctadecadienoic acid ((10R)-HPODE). Deletion of ppoA led to prominent reduction of (8R)-H(P)ODE and complete loss of (5S,8R)-DiHODE biosynthesis, whereas biosynthesis of (10R)-HPODE was unaffected. Deletion of ppoC caused biosynthesis of traces of racemic 10-HODE but did not affect the biosynthesis of other oxylipins. We conclude that ppoA of Aspergillus sp. may code for 5,8-LDS with catalytic similarities to 7,8-LDS and ppoC for linoleate (10R)-dioxygenases. Identification of these oxygenases and their products will provide tools for analyzing the biological impact of oxylipin biosynthesis in Aspergilli.

affect the biosynthesis of other oxylipins. We conclude that ppoA of Aspergillus sp. may code for 5,8-LDS with catalytic similarities to 7,8-LDS and ppoC for linoleate (10R)-dioxygenases. Identification of these oxygenases and their products will provide tools for analyzing the biological impact of oxylipin biosynthesis in Aspergilli.
The Aspergilli constitute a family of ascomycete fungi (1,2). Several species are important human allergens, opportunistic pathogens, and producers of mycotoxins. Their spores are ubiquitous in the environment. Immunocompromised patients are particularly vulnerable to infections by Aspergillus fumigatus, causing farmer's lung disease and invasive aspergillosis (1,2). Aspergilli are also plant pathogens and used as industrial microorganisms. Aspergillus nidulans is a model organism for studies of fungal biology (3). The genomes of nine Aspergillus sp. have now been fully or partly sequenced, which highlights their biological importance (2). One set of molecules known to be critical in Aspergillus developmental processes are a series of oxygenated fatty acids originally termed as psi 5

factors.
Champe and co-workers (4,5) showed in 1989 that A. nidulans oxidized 18:2n-6 and 18:1n-9 to psi factors, e.g. (8R)-HODE, (5S,8R)-DiHODE, and (8R)-HOME, which were identified as inducers of precocious sexual sporulation. Oxidation of polyunsaturated fatty acids to biologically active metabolites was well established in mammals and plants at that time, but this appears to be the first report of hormone-like activities of fungal oxylipins.
The biological importance of psi factors in the sporulation process of A. nidulans was further extended by Keller and coworkers (20 -23). The genomes of A. nidulans and A. fumigatus were published in 2005 (24,25). Keller and co-workers found with the aid of the 7,8-LDS sequence that both genomes contained three genes (ppoA, ppoB, and ppoC), which coded for putative fatty acid oxygenases of the MPO family with about 40% amino acid identity with 7,8-LDS (22). The exon-intron borders and the amino acid sequences of the gene transcripts could be deduced from sequence homology to 7,8-LDS, including homology to the presumed distal and proximal heme ligands of 7,8-LDS and the critical Tyr residue for catalysis. 6 The deduced sequence of PpoA of A. nidulans was confirmed by cDNA analysis (23). Keller and co-workers (20 -23) reported that deletion of these genes affected the ratio of asexual spores (conidia) to sexual spores (ascopores), the biosynthesis of (8R)-HODE, and mycotoxin production in A. nidulans. In addition to the impact on the sporulation process, deletion of these genes also led to alterations in virulence on host seed (20). Deletion of ppoA reduced formation of (8R)-HODE and increased the ratio of conidia to ascospores, whereas forced expression of ppoA had the opposite effect (23). Most recently, deletion of ppoB also increased conidia formation, whereas deletion of ppoC decreased conidia formation (20,21). These results were recently extended to A. fumigatus. An initial study demonstrated that down-regulation of all three A. fumigatus ppo genes by RNA interference technology produced a hypervirulent strain (27). Further work showed that deletion of ppoC yielded a pleiotrophic phenotype with formation of aberrant conidia and increased virulence in a mouse model of aspergillosis. 7 The biological effects of ppo gene loss in A. nidulans and A. fumigatus are summarized in Table 1.

Fungal Preparations
Fungal Growth-A. fumigatus and A. nidulans were grown in liquid media (1.5% malt extract) from spores or mycelia on agar in a rotary shaker (150 rpm) at 37 or 22°C (dark or in laboratory light) for 3-10 days. Mycelia were harvested by filtration, washed with saline, and either used directly or blotted dry and ground to a fine powder in liquid nitrogen. A. fumigatus and A. nidulans were also grown in 9-cm plastic Petri dishes either in the dark or 50 cm under a fluorescent lamp (30 watts, Tru-lite fluorescent, Duro-test, Fairfield, NJ, with or without light-dark cycles) for a few days at 22 or 37°C. Colonies were picked by forceps, blotted dry, and incubated with 18:2n-6.
Nitrogen Powder of A. fumigatus and A. nidulans-A. nidulans was grown in liquid culture for 3 days at 37°C (150 rpm), and A. fumigatus was grown for 24 h at 37°C and then at room temperature (150 rpm) for 48 h. Mycelia (10 -20 g) were harvested by filtration, washed with saline, and ground with liquid nitrogen to a fine powder, which was stored at Ϫ80°C. The nitrogen powder was homogenized (glass-Teflon, 10 passes; 4°C) in 10 volumes (w/v) of 0.1 mM KHPO 4 buffer (pH 7.3), 2 mM EDTA, 0.04% Tween 20, centrifuged at 13, 000 ϫ g (10 min, 4°C), and used immediately for enzyme assay.

Enzyme Assays
Incubation with Mycelia-Mycelia (0.5-20 g) were incubated with 5 volumes (w/v) of 0.1 M NaBO 3 buffer (pH 8.0 or 8.2) containing 18:2n-6 (0.5-1 mg/ml) for 5-6 h at 22°C with shaking. The pH of the incubation buffer was typically above pH 7.3 at the end. The mycelia were separated from the incubation medium by filtration. Medium from large scale incubations was extracted with ethyl acetate and from small scale incubations on SepPak/C 18 . The ethyl acetate extract was dried (Na 2 SO 4 ) and evaporated to dryness, and the products were purified by preparative TLC (ethyl acetate/hexane/acetic acid, 60:40:0.01) or by silicic acid chromatography. The latter was performed on a Sep/Pak cartridge or a short column with silica (Silicar CC-4, Mallinckrodt), which were eluted stepwise with increasing concentrations of diethyl ether (7, 25, and 50%) in hexane and finally with ethanol; 8-and 10-HPODE were eluted with 25% a The A. fumigatus ppoA, ppoB, and ppoC genes were identified as described previously (27). b The ppoA and ppoB knock outs were prepared by homologous recombination in the pyrG1 auxotrophic strain AF293.1 using A. parasiticus pyrG both as replacement cassette and marker gene, details in Footnote 7. c The ppoC knock out was prepared using the A. nidulans argB cassette and the pyrG1 and argB1 double auxotrophic strain AF293.6. 7 d TDWC10.5 was obtained by ectopic complementation with 5 kb of ppoC (containing 1 kb of its promoter 7 ). All mutants were characterized by PCR and Southern hybridization. 7 ether. The major metabolites were identified by LC-MS and by gas chromatography-MS analysis. Incubation with Subcellular Fractions-An aliquot (0.5 ml) of nitrogen powder supernatant was incubated with 30 -100 M of 18:2n-6, 30 -50 M [ 2 H]18:2n-6 or 15 M (8R)-HPODE for 30 -45 min on ice. The reaction was terminated with 0.5 ml of methanol, and the products were extracted on a cartridge of C 18 silica (SepPak/C 18 , Waters) as described (29). In some experiments, 50 pmol of (13R)-[ 2 H 4 ]HODE was added as an internal standard, and TPP (10 g) was added to reduce hydroperoxides to alcohols. The formation of (8R)-HODE and other oxylipins was quantified with help of the internal standard, (13R)- . Standard curves were prepared with 50 pmol of (13R)-[ 2 H 4 ]HODE and variable amounts of (8R)-HODE from a stock solution. The concentration of (8R)-HODE was determined by conversion to (8R,13R)-DiHODE by oxidation with manganese lipoxygenase (UV analysis) as described (29). Effects of drugs were assessed in duplicates with 100 M 18:2n-6 as substrate.
NP-HPLC-MS/MS was performed on silica with an analytical column (Kromasil-100SI; 250 ϫ 2 mm, 5 m, 100 Å), which was eluted with hexane/isopropyl alcohol/acetic acid, 95:5:0.05 or 93:7:0.95, using a Constametric 3200 pump (LDC). The effluent (0.3-0.7 ml/min) passed a photodiode array detector (5-cm path length, Surveyor PDA, Thermo Fisher Scientific). The effluent was then combined in a T junction with isopropyl alcohol/water (3:2; 0.3-0.5 ml/min) from a Surveyor MS pump and subjected to electrospray ionization in an ion trap mass spectrometer (LTQ, Thermo Fisher Scientific). The heated transfer capillary was set at 325°C, the ion isolation width at 1.5 atomic mass units, and the collision energy at 25 (arbitrary scale). Prostaglandin F 1␣ (100 ng per min) was infused for tuning. CP-HPLC analysis of 8-HODE and 10-HODE was performed as described (29). The effluent from the chiral columns was analyzed by MS/MS as described above.
The relative amounts of (8R,11S)-and (5S,8R)-DiHODE also appeared to be changed by temperature. The former was the main product at 22°C and the latter at 37°C (cf. Fig. 1, A and C).
We compared the transformation of (10R)-HPODE to 10-ODA by subcellular fractions of A. fumigatus with a heatinactivated control without detecting significant hydroperoxide lyase activity. The MS/MS spectra of 10-KODE and 10-ODA are shown in Fig. 3, A NOVEMBER 30, 2007 • VOLUME 282 • NUMBER 48
The reaction mechanism of the (8R)-and (10R)-DOX and the hydroperoxide isomerases were studied with stereospecifically deuterated 18:2n-6 as substrates for oxygenases/hydroperoxide isomerases of cell-free preparations of A. fumigatus. The results are summarized in Table 3.
These results suggest that biosynthesis of (8R)-HPODE occurs by abstraction of the pro-S hydrogen at C-8 and ant-

FIGURE 2. Biosynthesis of (8R)-HPODE and (10R)-HPODE by A. fumigatus.
A, LC-MS/MS analysis (m/z 311 3 full scan) of metabolites formed from 18:2n-6 by mycelia in 0.1 M NaBO 3 buffer. The amount of (8R)-HPODE varied from trace amounts to undetectable (cf. Fig. 1, A and B), but in some experiments (8R)-HPODE was obtained as a major metabolite. The MS/MS/MS spectrum (m/z 311 3 m/z 293 3 full scan) of (8R)-HPODE was as reported (35). B, demonstration of biosynthesis of (10R)-HPODE, 10-ODA, and 10-KODE from 18:2n-6 by nitrogen powder preparation of mycelia of A. fumigatus. The products were analyzed for 10-ODA (m/z 183 3 full scan; top trace) and for the carboxylate anions of (8R)-and (10R)-HPODE (m/z 311), as shown in the middle reconstructed ion chromatograms. Hydroperoxides decompose in the heated transfer system of the mass spectrometer to keto fatty acids. MS/MS analysis of KODE (m/z 293 3 full scan) showed that the major peaks of 8-KODE and 10-KODE co-eluted with (8R)-HPODE and (10R)-HPODE, respectively. Pre-formed 10-KODE and 8-KODE eluted before and after the two hydroperoxides. arafacial insertion of O 2 , whereas biosynthesis of (10R)-HPODE occurs by abstraction of the hydrogen at C-8, double bond migration to C-8 and C-9, and antarafacial dioxygenation at C-10 of the planar structure of C-8 to C-10. The hydroperoxidase activities, leading to hydroxylation reactions at C-5 and C-11, apparently occur by abstraction of the pro-S hydrogens at C-5 and C-11 of (8R)-HPODE and suprafacial insertion of oxygen.
Oxygenation of C 16 -C 20 Fatty Acids-The transformation of unsaturated C 16 -C 20 fatty acids by the fatty acid oxygenases of A. fumigatus is summarized in Table 4. (10R)-DOX apparently oxidized 16:1n-8, 18:1n-1(cis), 18:2n-6, and 18:3n-3. Several fatty acids were oxidized at their C-8 carbon and the corresponding hydroperoxide was apparently transformed to 5,8-and 8,11-diols. This seemed to require a saturated carbon chain from the carboxyl group to the first double bond of 7 or 9 carbons.

Effects of Deletion of ppoA and ppoC on Oxylipin Biosynthesis by A. fumigatus AF293
A. fumigatus AF293 formed (8R)-H(P)ODE, (5S,8R)-DiHODE, and (10R)-H(P)ODE as major metabolites. We could not detect biosynthesis of 8,11-DiHODE with certainty, but we cannot exclude that traces of this metabolite could be formed. Deletion of ppoA resulted in complete loss of biosynthesis of (5S,8R)-DiHODE, as judged from RP-HPLC analysis, and deletion of ppoC led to diminished biosynthesis of (10R)-HODE ( Table 5). Deletion of ppoB had little effect on oxylipin formation and transformed 18:2n-6 in the same way as A. fumigatus AF293.

Effects of Deletion of ppoA and ppoC of A. nidulans
The transformations of 18:2n-6 by the wild type and two mutants are summarized in Table 6.
Wild Type and ⌬ppoB-These two strain appeared to oxidize 18:2n-6 to the same spectrum of metabolites, e.g. (8R)-H(P)ODE, (10R)-H(P)ODE, and (5S,8R)-DiHODE, although the relative amounts differed. NP-HPLC of the hydroxy fatty acids formed by reduction with TPP did not reveal any qualitative difference in formation of hydroxy fatty acids between the wild type and ⌬ppoB. Biosynthesis of (8R,11S)-DiHODE could not be detected.
⌬ppoC-This strain had reduced the capacity to form (10R)-HODE (Table 6), and steric analysis of the small amounts of 10-HODE formed by this mutant showed that it was racemic (Fig.  7B). We conclude that ppoC codes for linoleate (10R)-DOX.

DISCUSSION
We have studied three previously uncharacterized fatty acid oxygenases of the human pathogen A. fumigatus and extended the results to the model organism, A. nidulans. (10R)-DOX and 5,8-LDS were found in both species and 8,11-LDS in A. fumiga-

TABLE 4 Oxygenation of unsaturated fatty acids by dioxygenases and hydroperoxide isomerases of A. fumigatus
The products were identified by MS/MS analysis of the carboxylate anions (A Ϫ 3 full scan) and by detection of characteristic fragments.
(10R)-HPODE partly decomposed to 10-ODA and 10-KODE during incubation (37), and it was also reduced to (10R)-HODE (Figs. 3B and 6). We could not detect transformation to diols. 10-ODA and 10-KODE were also formed during the LC-MS/MS analysis of (10R)-HPODE, presumably in the heated capillary. The transformation of (10R)-HPODE to 10-ODA appeared to be mainly nonenzymatic, as it was noted to a similar extent in heat-inactivated controls.
The genes of 5,8-LDS and (10R)-DOX were identified by gene targeting. Deletion of ppoA of both Aspergilli resulted in complete loss of biosynthesis of (5S,8R)-DiHODE and to biosynthesis of only small amounts of (8R)-H(P)ODE, whereas (10R)-HPODE now was formed as the major metabolite. Steric analysis showed that 8-HODE from these ppoA mutants contained moderate excess of the R stereoisomer. Deletion of ppoC resulted in almost complete loss of biosynthesis of (10R)-H(P)ODE, and the small amounts of 10-HODE formed was a racemic mixture of R and S stereoisomers. The changes in product formation in the ppoA and ppoC mutants (Tables 4 and 5) may not only be due to loss of the particular gene in question but also to a release of feedback inhibition observed by previous transcript analysis of ppo expression in ppo mutants (8,10). We conclude that ppoA codes for 5,8-LDS and ppoC for (10R)-DOX in both species. Table 7 summarizes these findings. The two 5,8-LDS enzymes can be aligned with 78% and the two (10R)-DOX enzymes with 66% amino acid identity. Whether 8-HODE can be formed by (10R)-DOX as a minor product or by other fungal enzymes awaits further studies.
Interestingly, the biosynthesis of (10R)-HPODE was augmented in both species grown at 22°C compared with 37°C. Deletion of ppoC in A. fumigatus augments the survival of this mutant in a mouse model of invasive aspergillosis. 7 Whether biosynthesis of (10R)-HPODE will affect the virulence of A. fumigatus may merit further investigation, as little is known about its virulence factors. In contrast, loss of ppoC had no effect on virulence in the A. nidulans/seed interaction. How-   ever, a double mutant deleted in both ppoC and ppoA showed a decrease in virulence on peanut seed (20). These results suggest possible different roles of oxylipins dependent on the host/fungal pathosystem.
Both A. nidulans and A. fumigatus contain a third gene (ppoB, see Table 2 and Table 7) that might code for fatty acid oxygenases. Alignment suggests that these genes may code for proteins with less than 40% amino acid identity and may not be closely related. Loss of ppoB had no discernible effect on virulence in A. fumigatus 7 but had in A. nidulans a large effect on increasing virulence on seed. 7 Disruption of ppoB did not influence oxylipin biosynthesis in this study. Recent mRNA analysis shows A. nidulans ppoB to be a pathogenesis-induced gene, 9 which may explain why changes in oxylipin profile were not observed in this strain in the present study. A. fumigatus AF293 may only produce traces of (8R,11S)-DiHODE, whereas A. fumigatus Fres. formed this metabolite as one of the major products. In analogy, Aspergillus clavatus contains three genes with homology to the three oxygenase genes of A. fumigatus, and this fungus also expresses 5,8-LDS, 8,11-LDS, and (10R)-DOX activity. 10 It is possible that the PpoB of A. fumigatus forms 8,11-LDS, a topic of further studies.
The MPO family contains fatty acid dioxygenases with hydrogen abstraction by a tyrosyl radical as a common feature. The homology of oxygenases of A. nidulans and A. fumigatus include the distal and proximal heme His ligands and the catalytically important Tyr residue of cyclooxygenases. In agreement with this oxygenation mechanism, TNM (30 -100 M) inhibited the (8R)-and (10R)-DOX activities of A. fumigatus, possibly by interfering with the oxygenation mechanism by nitration of Tyr residues.  What are the structural differences between (8R)-DOX and (10R)-DOX? As discussed above, a Tyr radical formed by both groups of enzymes likely abstracts the pro-S hydrogen at C-8 of 18:2n-6, but oxygen then reacts either at C-8 or at C-10 of 18:2n-6. It is known from lipoxygenase biochemistry that mutation of a single amino acid can change the position of oxygenation (26). It is therefore of interest to determine the conserved differences in the primary sequences of 5,8-LDS and (10R)-DOX and to compare them with homologous positions in cyclooxyenases. Replacement of Ser-530 of cyclooxygenase-1 with threonine or acetylation of the corresponding Ser residue of cyclooxygenase-2 with aspirin shifted the position of oxygenation of 20:4n-6 from C-11 to C-15 (28). Cyclooxygenase-1 has conserved Tyr-348 and Val-349 residues, which are important for substrate positioning and for the cyclooxygenase reaction (28). Replacement of Val-349 with a larger (leucine) or smaller residue (alanine) increased the oxygenation at C-15 and C-11, respectively (28). There are putative oxygenases with close similarity to PpoA (5, (26,28). It will be of interest to study the effect of replacement with a larger residue, e.g. V330L.
In summary, we have identified novel oxygenases and oxylipins in two Aspergilli sp. Our results indicate that diol synthases and (10R)-DOX have fundamental catalytic and structural properties in common. The genes of these diol synthases and (10R)-DOX of A. nidulans and A. fumigatus have been deleted, and the resulting phenotypes can now be interpreted in the light of our report.